What we’re building
We are developing a network of smart bat monitors, each one capable of listening to the environment, deciding if bats are present and then figuring out what species they are – think of each smart bat monitor as Shazam for bats. To create the smart bat monitors we are using the Intel Edison with Arduino breakout, plus the Dodotronic Ultramic 192K microphone.
What each device will do
First – a microphone on each device, capable of handling ultrasonic frequencies, will listen to all audio from the environment up to 96kHz. Most bats talk at frequencies above 20kHz (the limit of human hearing) with some species going as high as 125kHz (although none of these species are found in the Queen Elizabeth Olympic Park).
Second – signal processing techniques will be used to detect if any loud noises (such as bat calls) are occurring in the ultrasonic frequencies above 20kHz. If so, the device will begin to record the audio from the environment, and keep recording until the loud noises at these ultrasonic frequencies stop.
Third – the recorded audio is then turned into a spectrogram image using a method called Fast Fourier Transform. The spectrogram image shows the amplitude of sounds across the different frequencies over time. Bat calls can clearly be seen on the specrogram as bright patterns (indicating a loud noise) at high frequencies.
Finally – image processing techniques, called Convolutional Neural Networks (CNN), are applied to the spectrogram images to look for patterns that resemble bat calls. If any suspected bat calls are found in the image, then the same CNN techniques are applied again to each individual bat call to look at its shape in more detail and determine what species of bat it most likely is.
Where we are monitoring bats
We are aiming to deploy a network of up to 15 smart bat monitors across the Queen Elizabeth Olympic Park in early 2017. Over several months we will monitor their performance as well as investigating levels of bat activity and what species are present in the park.